From: Adam M. C. <png...@ni...> - 2005-03-20 12:29:19
|
John Denker <js...@av...> wrote: > 2) If you pick a reference point at x,y = 1/3, 1,3 let's call > that your "W" point. That's fine in the abstract, and sometimes > abstraction is good. But let's not call it the WHITE point, because > under many conditions it would not be perceived as white. There is > some latitude as to what should be called WHITE, but not unlimited > latitude. Sure, but by the same argument, we shouldn't call the three x,y pairs "red", "green", and "blue" chromaticities either. > My point here in item (2) is that most people -- and most tools -- > desire and/or assume that R,G,B = 1,1,1 in document space corresponds > to something the *looks* white. So my xRGB system is designed to > accomodate these desires. There is a similar assumption that 1,0,0 looks red, 0,1,0 looks green, and 0,0,1 looks blue. But in xRGB (and in XYZ) those points don't "look" like anything at all, because they correspond to imaginary situations where negative light is striking your retina and producing a negative response in some of the cones. By adopting such wacky chromaticity values, we've already strayed outside the realm of physical reality, and violated some basic assumptions. Once we've gone that far, I figure we might as well go all the way, do the same with the white point, and end up with a color space that's already extremely well known and standardized, rather than define yet another one. > We agree the chromaticity vector is related to the tristimulus vector > in the simple way > [x,y,z] = [X,Y,Z] / (X+Y+Z) Yes. > What are the proper name and symbol for (X+Y+Z), the denominator in > this expression? I don't know. Does anyone else know? Does X+Y+Z have its own name and symbol? > I don't think anybody would perceive [X,Y,Z] = [1,0,1] You're correct if you stop the sentence right there. :) (As John already pointed out.) > X Y Z R G B > 0.039 0.077 0.014 0.000 0.107 0.001 > 0.183 0.077 0.950 0.000 0.007 0.999 I converted those XYZ values to sRGB and obtained: P3 1 2 255 1 83 2 1 8 255 For anyone who doesn't read PPM format, that's two pixels whose RGB values appear on the last two lines. When I displayed this on my CRT and scaled it up, the blue patch looked just slightly brighter than the green patch, but this monitor is almost certainly not sRGB. (It's borrowed. When I first got it, the colors looked terrible, so I played with the individual R,G,B settings until they looked non-terrible, but I didn't do any real calibration.) > And FWIW, in an abundance of thoroughness, I fired up > ghostscript and compared > 0.000 0.107 0.001 setrgbcolor > against > 0.000 0.007 0.999 setrgbcolor I have no idea whether the PostScript setrgbcolor function expects gamma-encoded values. AMC |